arXiv:2604.02006v1 Announce Type: new Abstract: Reinforcement Learning (RL) significantly enhances the reasoning abilities of large language models (LLMs), yet applying it to multi-turn agentic tasks remains challenging due to the long-horizon nature of interactions and the stochasticity of environmental feedback. We identify a structural failure mode in agentic exploration: suboptimal actions elicit noisy observations into misleading contexts, which further weaken subsequent decision-making, making recovery increasingly difficult. This cumulative feedback loop of errors renders standard exploration strategies ineffective and susceptible to the model's reasoning and the environment's randomness. To mitigate this issue, we propose ProCeedRL: Process Critic with Explorative Demonstration RL, shifting exploration from passive selection to active intervention. ProCeedRL employs a process-level critic to monitor interactions in real time, incorporating reflection-based demonstrations to guide agents in stopping the accumulation of errors. We find that this approach significantly exceeds the model's saturated exploration performance, demonstrating substantial exploratory benefits. By learning from exploratory demonstrations and on-policy samples, ProCeedRL significantly improves exploration efficiency and achieves superior performance on complex deep search and embodied tasks.

arXiv:2511.18123v2 Announce Type: replace-cross Abstract: Vision-Language Models (VLMs) have become indispensable for multimodal reasoning, yet their representations often encode and amplify demographic biases, resulting in biased associations and misaligned predictions in downstream tasks. Such behavior undermines fairness and distorts the intended alignment between vision and language. Recent post-hoc approaches attempt to mitigate bias by replacing the most attribute-correlated embedding coordinates with neutral values. However, our systematic analysis reveals three critical limitations of this coordinate-wise approach: feature entanglement, poor cross-dataset generalization, and incomplete bias removal. We find that bias is not localized to a few coordinates but is instead distributed across a few linear subspaces. To address these limitations, we propose $\textbf{S}$ubspace $\textbf{P}$rojection $\textbf{D}$ebiasing ($\textbf{SPD}$), a geometrically principled framework that identifies and removes the entire subspace of linearly decodable bias while reinserting a neutral mean component to preserve semantic fidelity. Extensive experiments across zero-shot classification, text-to-image retrieval, and image generation validate the effectiveness of SPD: our method achieves more robust debiasing with an average improvement of $18.5\%$ across four fairness metrics, while maintaining minimal loss in task performance compared to the best debiasing baseline.

Background: During the diagnosis and treatment of non–small cell lung cancer (NSCLC), detecting the risk of its recurrence in an early phase is still challenging. Recent studies have investigated the radiomics-based machine learning (ML) models for detecting the risk of recurrence in NSCLC. However, there is still insufficient systematic evidence to prove its efficiency. Objective: This study is designed to systematically evaluate the effectiveness of radiomics-based ML in predicting the risk of recurrence in NSCLC, aiming to provide evidence-based support for the subsequent development of scoring tools to forecast recurrence risk. Methods: For acquiring research on radiomics-based models for forecasting the risk of recurrence in NSCLC, Cochrane Library, Web of Science, PubMed, and Embase were systematically retrieved, up to October 24, 2025. Studies on analyzing the recurrence of NSCLC using radiomics-based ML were included, while those in which only texture analysis was conducted or radiomics-based ML was not constructed were excluded. The Radiomics Quality Score (RQS) was used to appraise the eligible studies. Subgroup analyses were conducted according to the variables of the model, the background of treatment, the stage of lung cancer, and the pathological type. Results: Ultimately, 30 eligible studies in total were included, covering 7964 patients with NSCLC. According to the meta-analysis, the c-index of radiomics-based ML models for forecasting the risk of recurrence in NSCLC was 0.850 (95% CI 0.834‐0.866, 95% prediction interval [PI] 0.623‐1.004) in the training set. Specifically, the pooled c-index was 0.876 (95% CI 0.853‐0.900) among the patients receiving the stereotactic body radiation therapy and 0.825 (95% CI 0.804‐0.848) among those who received surgeries combined with other adjuvant treatment regimens. The c-index of the radiomics-based ML models combined with clinical features for forecasting the risk of recurrence in NSCLC was 0.833 (95% CI 0.822‐0.854, 95% PI 0.717‐0.945) in the training set. In contrast, the c-index of radiomics-based ML models for forecasting the risk of recurrence in NSCLC was 0.878 (95% CI 0.854‐0.902, 95% PI 0.681‐1.000) in the validation set. The c-index of radiomics-based ML models combined with clinical features for forecasting the risk of recurrence in NSCLC was 0.854 (95% CI 0.830‐0.878, 95% PI 0.655‐0.992) in the validation set. The average RQS across the included studies was 27.4%, revealing methodological limitations and an absence of standardization. Conclusions: This study is the first to confirm that radiomics-based ML models effectively predict the risk of recurrence in NSCLC. This study provides evidence-based support for the subsequent development or updating of radiomics-based ML models. However, the current methodological application of radiomics remains concerning. Therefore, in the future, research should standardize the workflow for implementing radiomics-based ML and incorporate multicenter imaging data to enhance its generalizability. Trial Registration: PROSPERO CRD42025631191; https://www.crd.york.ac.uk/PROSPERO/view/CRD42025631191